Subject 8

Process Synthesis:
Heat Exchanger Network Synthesis
(HENS)
Javier R. Viguri / Eva Cifrian
[email protected] / [email protected]

Department of Chemistry and Process & Resource Engineering

GER Green Engineering and Resources Research Group
www.geruc.es
1
INDEX

1.- Heat Exchanger Network Synthesis (HENS): Introduction

2.- HENS: Minimum Utility Consumption
3.- HENS: Minimum Number of Units
4.- Optimum approach temperature
5.- Heat Integration
6.- Further Readings and References

PRACTICAL CHAPTER
RELEVANT TO LEARNING

2
 A

1.- Heat Exchanger Network Synthesis: Introduction

Objective: Recover Heat to reduce energy consumption. V2

RE 9 10

SP1

6
H2
CL C3 V1
M3 R2 12 EX1 EX2 5 SS1
EX3 C1 R1 4
EL
3

C1 34
H1
78
H4 H3
HCL 8

23
7 D2 44
C2 D1
C4
C2 EX4
88 66

C3
VCL

Cool Stream → C# Streams that need steam. Sinks heat

Hot Stream → H# Streams that need cooling water. Sources Heat
Basic HENS problem
• The Mass and heat Balances have been completed
• Stream characteristics: Tinlet, Toutlet, heat capacity flowrate
f (KW/K) = F (kg/s) x Cp (J/kg.K) ; Q (KW) = f Tp
• Non-isothermal streams Tinlet ≠ Toutlet (no change of phase)
• Change of phase.
• Pure components Tinlet = Toutlet isothermal streams (vaporize, condense).
Assumptions: Tp = 1; Q = fTp; f = H 3
• Multicomponents Tp =Tdew-Tbub → Q = fTp; f = H/Tp
1.- Heat Exchanger Network Synthesis: Introduction
HEN Problem

Given nH hot streams and nC cold streams

* HOT fHi, THiin, THiout
+ Auxiliary utilities such as fuel, steam, cooling water, refrigeration

* COLD fCj, TCjin, TCjout

Find the network of countercurrent heat exchangers (single pass) to

MINIMIZE total annualized Costs CA (exchangers +utilities)

 
C A = im (CCTI ) + C = C A = im  C P , I i +  C P , A j  + sFS + (cw )Fcw
 i j 
im: return of investment annually (0.33);
CCTI: Total Capital Investment;
C: Annual Cost of utilities;
CP,Ii: Purchase cost of exchangers;
CP,Aj: Purchase cost of auxiliary network;
4
Fs, Fcw: Annual flowrates of steam and cooling water; s, cw: unit cost of steam and cooling water
1.- Heat Exchanger Network Synthesis: Introduction
H1OUT

Alternatives (three): (1.- Extreme, 2.- Near Optimal, 3.- Optimal)

1.- Extreme Alternative
H1IN CT

a) Every hot stream with cooling utility → Cooling Towers

Every cold stream with heating utility → Generating steam

C1IN GS
Higher driving force → U↑ →A↓

→ Is the minimum possible interchange between streams

b) Recover as much heat as possible using the least amount of utilities

H1 H1 H2 C1H
H2

C1 1 2

C23 H1C
C2
H13
C2 ST
3

Optimal solution from a heat recovery point of view because only need heating (or
5
cooling); This is the maximum possible heat interchange. Not always possible.
1.- Heat Exchanger Network Synthesis: Introduction
Alternatives.
2.- Near Optimal Network (two-step synthesis)

a) Minimum Utility Cost (consumption) for a given Tmin (HRAT-heat recovery

approach temperature) → Minimize Utility Costs e.g. HRAT= 10 K
b) Fewest number of units (exchangers) The Mass and heat Balances have been
completed → Minimize investment Costs

Pinch Analysis → 1.- Predict a) + b); 2) Develop network structure to satisfy a) + b)

70´s. Linnhoff. Linde and other
separation air industries use this
methodology http://www.madehow
.com/Volume-
4/Oxygen.html

3.- Optimal Network (one-step synthesis)

MINLP Optimization of CA → Non Lineal € =  (S),  < 1; C =Co (S/So)

Tlm= (T1-T2) / ln (T1/T2) 6
1.- Heat Exchanger Network Synthesis: Introduction
2.- Near Optimal Network (two-step synthesis)

Stream, Utility data, Constraints, Mass and Heat balance

Strategy
Minimization of utilities (Min Utility Cost) effective when
Temperature Interval Method €fuel > €equip
Graphical Method
Cost

Linear program (LP)

Utilities Cost

Stream Matching
At the Pinch
Exchangers

Mixed Integer linear program (MILP)

Adjust
Tmin
Fewest number of units (Min Investment Cost)
Break Heat Loops
Introduce Stream Splitting

Calculate CA

Final Network 7
2.- HENS: Minimum Utility Consumption
At specified Tmin= HRAT
f (KW/K) Tin (K) Tout (K) First Law
H1 10 600 320 7200 Kw Thermodynamics:
H2 20 540 320 7200 – 5550 Kw =
C1 15 360 600 1650 Kw cooling
5550 Kw
C2 13 300 450

Pinch Analysis: Use of the composite Streams (Hohmann Curves) to

obtain HEN by Graphical Approach. Temperature interval method.

* At HRAT specified → Q = fTp → Tp = Q/f → Slope 1/f

You specify that:
tmin = HRAT = 0 → Area = ∞; Tmin= HRAT ↑↑ → Energy ↑↑ and A → 0
•Influence of the “Minimum Recovery Approach Temperature-HRAT” or
“Approach Temperature-Tmin” Tmin= HRAT

* Problem Table (Heat flow analysis) at HRAT=20 K. Temperature intervals

in order to guarantee feasible heat exchange based on the inlet Ts.

8
* Final result from the Graphical and Interval approaches are the same
 f (kW/K) Tin(K) Tout(K)
H1 10 600 320
Pinch Analysis
H2 20 540 320
C1 15 360 600
C2 13 300 450

T
Heating
600
1/10 H1

Composite Hot
H1 + H2 1/15 C1
500
1/30

Composite Cold
400
1/28 C1+C2

1/13 C2
300

Cooling Heat recovery

Q
 Target min heating/min cooling

Min Heating 600KW

600
H1

HRAT=20K PINCH!! (540-520)

Composite Hot C1
500 H1 + H2

Composite Cold
400
C1+C2

C2
300
Min Cooling 2250kW

10
 Problem Table Tmin=HRAT=20K Feasible
Cascade
Temperature intervals
Hot Cold Heat Surplus/ Infeasible
= Cold + HRAT = Cold
Deficit kW Cascade 600 kW min heating
(620) 600

H1
-300 -300 300
600 (580)

H2 -300 -600 0 PINCH!

540 (520)
450
1230 630 1230
(380) 360
C1 1020 2250
1650
(320) 300
10 20 C2
15 13
2250 Min cooling

f (kW/K) Tin(K) Tout(K) Min heating 600 kW

H1 10 600 320
Min cooling 2250 kW
Pinch 540-520K
H2 20 540 320
C1 15 360 600
C2 13 300 450

11
Heat Exchanger Network Synthesis

The Problem Table (Heat flow Analysis)

1.- Determine Heat Surplus/Deficit in each interval

2.- Cascade heat with 0 Kw heating. Largest Deficit = Min Heating
3.- Cascade with Minimum Heating
3.1. Min Cooling at bottom
3.2. Temperature below position where 0 Kw entering interval → PINCH.

Not heat transfer below → Partition of Design between ABOVE + BELOW

pinch

12
Mathematical Optimization Models for HEN: Transshipment model
(Papoulias and Grossmann, 1983)

Heat Content at HRAT = 20 K

Temperature intervals Heat Content (kW)

Hot Cold H1 H2 C1 C2
(620) 600

H1
300
600 (580)

H2 600 900
540 (520)
450
1600 3200 2400 1170
(380) 360
C1 600 1200 780
(320) 300 15
10 20 C2 2800 4400 3600 1950
13

13
Mathematical Optimization Models for HEN: Transshipment model
(Papoulias and Grossmann, 1983)

Heat Cascade Diagram at HRAT = 20 K

Sources QH 125 €/kW yr Sinks

620 600
1

600
R1 580 300
2800 600 3600
kW H1 2 900 C1 kW
1600 2400
600 540
R2 520
4400 H2 3200 3 1170 1950
kW C2 kW
380
R3 360 780
1200
4
320 300

QCW 25 €/kW yr
14
Mathematical Optimization Models for HEN: Transshipment model
(Papoulias and Grossmann, 1983)
Heat Balances
R1 + 300 = QH
R2 + 900 = R1 + 600 4 eqs; 5 var R1, R2, R3, QH, QCW
R3 + 3570 = R2 + 4800
QCW + 780 = R3 + 1800

Linear programming problem (LP)

Min Z = 125 QH + 25 QCW

s.t. R1 - QH = - 300
R2 – R1 = -300
R3 – R2 = 1230
QCW – R3 = 1020 R1, R2, R3, QH, QCW ≥ 0

Solution

QH =600 kW; QCW =2250 kw; R1=300; R2 = 0 → PINCH (540-520); R3=1230; Z= 131250 €/yr
600 580
Remarks:
2
1.- The approach can be easily extended to multiple utilities
560 540
(e.g. use of steam at 560 K) by simply including in heat balance. Qst
New
540 520
2.- Can extend to forbidden matches (e.g. No H1-C1)
15
3.- HENS: Minimum Number of Units. Stream matching at the pinch

For each subnetwork (Above / below the pinch) assume at least one stream is
exhausted in each match:
Hot streams / Utilities Given a graph with n nodes

Cold Streams / Utilities the minimum links = n-1

For each subnetwork: Nmin = ncold + nhot + nutility -1

In our Example we can predict the nº of exchangers:
Above pinch: H1, C1, Steam → Nmin = 1 + 1 + 1 – 1 = 2 units
Below pinch: H1, H2, C1, C2, Cooling Water → Nmin = 2 + 2 + 1 – 1 = 4 units
Total Nmin = 2 + 4 = 6 units
QH = 600 kW heating at 125 €/kW yr;
QCW = 2250 kw cooling at 25 €/kW yr;
Pinch: 540-520 K;
€ = 131.250 €/yr to HRAT = 20 K
16
3.- HENS: Minimum Number of Units. Stream matching at the pinch

Derivation of the Network (Non-systematic. Only optimization is systematic)

Above pinch (Only H1, C1, Stream) Check if the

(620) 600
15 interchange is feasible
(10) H1
H1
600 (580)
600 600 540
520 560 600 540 Pinch (520) H1
H11
10 C1 EX1
C1 C11 C1H C1
1 H C11
(15) 520
560
H11 Steam
600 kw
540 (600-540)x10=600 = (T-520)x15 → T=560

Below pinch (H1, H2, C1, C2, CW)

H1-C1 Infeasible Check if the
H1
H2-C1, H1-C2 (20) H2

540 540 interchange

(10) is feasible
C1 C12 C2 C23
H1H2 2 3
540 Pinch(520) 360 520 300 450
450 540 345
H22 H13 H13
H1
420 345 EX3
(380) 360 C2
C23
Cooling Cooling 300
C1 450
2000 kw H C 250 kw
(320) 300
10 20 15 C2 H2C H1C
13 320 320 17
3.- HENS: Minimum Number of Units. Stream matching at the pinch

Final Network Structure 540

H2
H1 600

360
520 560 600
C1

2 1 H

420
540 Steam
H22

Consider 600 kw
C12

CW2
C2
450
300 3

Cooling
Uexchangers= 0.8 kw/m2K 2000 kw
320
345
H2CW2

Uheaters= 1.2 kw/m2K

Total Area = 217 m2 CW1
Cooling 250
Investment Cost = 72,960 €/yr 320 kw

Cost Exchangers= 1,500 A0.6 €/y

Utility Cost = 131,250 €/y

TOTAL: 204,210 €/y

By Optimization the TOTAL: 184,750 €/y

Not too far because the utilities are low

18
3.- HENS: Minimum Number of Units. Grid Diagram

Representation of the HEN, with stream, f, T, pinch, units (exchangers and

utilities), heat interchanged, Matches. 540
H2
H1 600

360
520 560 s 600
C1
2 1 t
1.- Diagram 2

420
1

540
H

Steam
H22
C11
c 600 kw
C12

2.- Matches 1, 2, 3 w
3 450
CW2
C2

300 3 H11 C23

3.- Utilities Cooling

2000 kw
320
345
H2CW2

c
4.- T and Kw
H13

w CW1
Cooling 250
320 kw

Q= 1950 kw Cooling
HOT Q= 600 kw Pinch 250 kw COLD
600 540 540 345 320
H1 (10) 1 3 cw

Q= 2400 kw
540 540 540 420 320
H2 (20) 2 cw

Cooling
600 520 520 2000 kw 360
560
st 1 2 C1 (15)

Steam Q= 600 kw Q= 2400 kw

600 kw 520 520 450 300
3 C2 (13)

Q= 1950 kw
19
3.- HENS: Minimum Number of Units. Stream matching at the pinch

540
Optimum Network Structure H2
H1 600

360
530 574 600
The same structure as the C1

previous one 2 1 H

412
534.4 Steam
H22
C12
388 kw
Consider CW2
C2
450
Uexchangers= 0.8 kw/m2K Cooling
300 3

320 H2CW2
340.7
Uheaters= 1.2 kw/m2K 1843 kw

Total Area = 238 m2 CW1

Cooling 194
Investment Cost = 72,960 €/yr 320 kw

Cost Exchangers= 1,500 A0.6 €/y

Exchanger Cost: 87,237 €/y
Utility Cost = 97,515 €/y

TOTAL (optimization): 184,750 €/y vs. TOTAL (Pinch) 204,210 €/y

Same Structure of units but different temperatures 20

3.- Automatic Synthesis at HENS: Mathematical Optimization Models for
HEN (Papoulias and Grossmann, 1983 b,c).

Objective

Simultaneous Optimization at: Heat Recovery + Selection of matches

to minimize Total Annual Costs → Investment Costs Exchangers + Utility Costs

Procedure
1.- Postulate Staged Superstructure at alternative configurations
2.- Model as a Mixed-Integer NonLinear Program (MINLP)
3.- Determine solution by solving the MINLP problem

H1 Each exchanger may be selected or

not (0-1).
Specific configurations are obtained
by finding values at 0-1 variables,
C1
plus heat loads and temperatures.

21
3.- Mathematical Optimization Models for HEN: MINLP model
Cold j

Hot i
zijk : 1 selected; 0 not selected
zijk , qijk qijk : Heat load
tik-1 tik tik : T stage k

Staje k

min Z = Fixed Cost exchangers + Cost areas + Cost utilities

s.t. heat balances for energy stream, each stage monotonic decrease T (left → right)

Logic constraints:

a) If zijk = 0 → qijk = 0 linear ineq. Qijk ≤ Ω zijk , qijk ≥ 0

b) If zijk = 1 → positive driving force tik – tjk ≥ Tmin – M (1- zijk)

Note:
1.- Need to specify number of stages (e.g. nº stages = max {ºn hot, nº cold} )
2.- Nonconvex MINLP → can be trapped in local solutions. To avoid bad solutions can add
constraint → Heating utility ≤ Qtarget,HRAT; Qtarget, HRAT from LP transshipment model
Model implemented in PC based software (SYNHEAT):
- automatically interfaces with GAMS
- includes LP transshipment 22
4.- Optimum “Minimum Recovery Approach Temperature-HRAT” or “Approach
Temperature-Tmin” (Mínima diferencia de temperatura permitida)

Tmin →0 A→ ∞
Utility requirements → Absolute minimum

Tmin → ∞ A→ 0
Utility requirements → Maximum (not heat exchange between
process stream).

Ttreshold Critical Tmin below which no pinch exists (Threshold Approach T Difference)
Costs

Utilities

Capital
Ttreshold

Tmin
Designing HEN → Consider effect of Tmin 23
5.- Heat integration

Flows and temperatures in a process can be adjusted to improve heat recovery

5.1.- Readjusting pressures and temperatures

5.1.1. Distillation columns (highly energy-intensive)
5.1.2. Multiple-effect distillation
5.2. Readjusting flows
5.2.1 Recycles

5.1.1. Readjustment P´s in distillation columns

Assume heat load similar in Condenser and in reboiler → Qcooling≈ Q steam

(reasonable approximation if saturated liquids and Tcond < TF < Treb)

Heat Source
D Qsteam
Qcooling Treb

D T
F
Qcolumn
Tcond
Heat Sink Qcooling
B
Qsteam Q
24
5.1. Readjustment P´s in distillation columns

The heat duty Q ≈ Qcooling≈ Q steam is related directly to the reflux ratio. When Q is
reduced (high cost of fuel) the Nº trays increases →
→ Tradeoffs {Operating costs vs. Capital Costs}.
Qsteam
Heat integration
a) Adjustment of the P level of the column to T Q

Treb

position its T-Q rectangle to lie below the hot Q

Tcond
Q
composite and above the cold composite curves.

Qsteam
Q

Treb
Qcw
T Q
Tcond Q
Q

b) The P of the column is adjusted and the column

positioned so that the reboiler receives its energy
from a hot utility and the condenser rejects its
Qcw energy to the cold process streams above the pinch.
25
Q
5.1.2. Multiple-effect distillation

Separations where T-Q rectangle cannot be positioned under hot composite

curves and above cold composite curves → Multiple effect distillation.

D1
D
I M1
Low P D

F1
F D
D2

F S1

F2

B1
F3 M2
B
B II
High P
P1

Qeffect B2

Treb,2
50% reduction in steam but higher investment + higher
II
Qsteam grade steam
Tcond,2
Tmin
P ↑ in (Neff-1) columns → steam cost ↓ (Qeff ≈ Q/Neff)
Treb,1
T
Qcolumn I ↑ € pumping
↑ € equipment (thickness)
Tcond,1 ↑ € ↓→nº trays↑ → H↑ vs. ↓ € steam
Q cooling Qeffect ↑ € higher grade steam
↑ € high investment 26
Q
5.2. Readjusting flows

5.2.1 Recycles

During heat integration it is possible increase recycle and save on raw materials
10
Total

Costs
7 C2
9
no heat
 SP1 integration
Total
with heat integration
6

EX2
XIX1 EX1 SS1
C1 R1

1
2
5 Raw materials

8

1.- With heat integration → curve shifted to right to higher  → higher overall
conversion.
2.- Trade-off: Energy Cost vs. raw material Cost

Procedure
1.- Fix 
2.- Calculate mass + heat balance
3.- Determine minimum utility cost
4.- Total Cost = Raw material + Compression + minimum utility cost
5.- Repeat for different values at  to select minimum cost solution. 27
6.- Further Reading and References

• Biegler, L., Grossmann, I., Westerberg , A., 1997, Systematic Methods of Chemical
Process Design, Prentice Hall.

• Kemp, I., 2007, Pinch Analysis and Process Integration. 2nd Ed. Elsevier.

• Papoulias, S., Grossmann, I.E., 1983a, A structural optimization approach in Process

Synthesis-I: Utility Systems. Computer and Chemical Engineering, 7, 695.

• Papoulias, S., Grossmann, I.E., 1983b, A structural optimization approach in Process

Synthesis-I: Heat recovery network. Computer and Chemical Engineering, 7, 707.

• Papoulias, S., Grossmann, I.E., 1983c, A structural optimization approach in Process

Synthesis-III: Total Processing Systems. Computer and Chemical Engineering, 7, 723.

•Ravagnani, M., Caballero, J.A., Redes de cambiadores de calor. Publicaciones de la

Universidad de Alicante. 2012.

• Sinnot, R, Towler, G., 2009, Chemical Engineering Design. 5th Ed. Coulson &
Richardson´s Chemical Engineering Series. Elsevier.

• Seider, W., Seader, J., Lewin, D., Widagdo, S., 2010, Product and Process Design
Principles. Synthesis, Analysis and Evaluation. 3rd Ed. John Wiley & Sons.

28
PRACTICAL CHAPTER

• Examples of HEN: Composite curve, Cascade of

temperatures.

29
 RELEVANT TO LEARNING (I)
➢ Where are the sources and sinks of heat in a chemical process?
➢ Heat Exchange Network (HEN)
- What is the Cost?
- Alternatives to doing HEN synthesis
- Algorithm to obtain a network
➢ Obtain the PINCH and Min. Heating and Cooling needs by means of:
- Composite curves. Dada una representación T vs. H indicar toda la información que se
puede extraer respecto del diseño de una red de intercambio de calor.

- Feasible cascade

➢ To draw a grid diagram of a HEN

➢ Meaning of Tmin

➢ How to do heat integration in distillation columns

30
 RELEVANT TO LEARNING (II)

➢Given a set of hot and cold streams, assume that the result obtained
from a heat exchange network, working at 10K HRAT, is a prediction of
minimum services of 1800 kW of heating and of 2400 kW of cooling.
Assume that HRAT is changed to a value of 20K and then 2200 kW of
heating and 2600 kW of cooling are predicted. Determine whether or not
these results make sense and why.

➢Dado un conjunto de corrientes calientes y frías, suponer que se

obtiene como resultados de una red de intercambio de calor, trabajando a
HRAT de 10K, una predicción de unos mínimos servicios de 1800 kW de
calentamiento y de 2400 kW de enfriamiento. Suponer que se cambia
HRAT a un valor de 20K y entonces se predice 2200 kW de calentamiento
y 2600 kW de enfriamiento. Determinar si estos resultados tienen sentido
o no.

31
 RELEVANT TO LEARNING (II)

➢In a heat exchange network there may be more than two pinch points.
Explain this briefly.

➢¿En una red de Intercambio de calor pueden existir más de dos puntos
pinch?. Explicarlo brevemente.

---------------------------------------------------------------------------------------------------------

➢A two-stage strategy for designing and optimizing a heat exchange

network is better than a single-stage global optimization strategy.
Discuss.

➢Una estrategia de diseño y optimización de una red de intercambio de

calor en dos etapas es mejor que una estrategia de optimización global
en una sola etapa.

32